Stepwise Variable Selection returns to the stepwise control panel.
Discriminant Method chooses the discriminant method. Details are shown in the section Discriminant Method.
Score Data shows or hides the listing of the scores by row in the Discriminant Scores portion of the report.
Score Options deal with the scoring of the observations and includes the following:
shows rows that are misclassified and those where p>.05 and p<0.95) for any p, the attributed probability.
saves formulas to the data table. The distance formulas are Dist[0], needed in the Mahalanobis distance calculations, and a Dist[ ] column for each X-level’s Mahalanobis distance. Probability formulas are Prob[0], the sum of the exponentials of -0.5 times the Mahalanobis distances, and a Prob[ ] column for each X-level’s posterior probability of being in that category. The column includes a Response Probability property. The Pred column holds the most likely level for each row.
Canonical Plot shows or hides the Canonical Plot.
Canonical Options provides commands that affect the Canonical Plot and include the following:
colors the points based on levels of the X variable. This statement is equivalent to selecting Rows > Color or Mark by Column and selecting the X variable.
shows or hides the Canonical details. Details for the Iris data set are shown in Canonical Details. The matrices at the bottom of the report are opened by clicking on the disclosure icon beside their name and closed by clicking on the name of the matrix.
Canonical Details
Canonical 3D Plot is available only when there are four or more groups (levels). An example of this plot is shown in Canonical 3D Plot. It shows a three-dimensional version of the Canonical Plot and respects other Canonical Options.
Canonical 3D Plot
1.
Open cereal.jmp.
2.
Click on Analyze > Multivariate Methods > Discriminant.
3.
Specify Calories, Protein, Fat, and Sodium as Y, Covariates.
4.
Specify Manufacturer as X, Categories and click OK.
Specify Priors lets you specify the prior probabilities for each level of the X variable:
Specify Prior Probabilities Dialog
Consider New Levels is used when you have some points that may not fit any known group, but instead may be from an unscored, new group.
Save Discrim Matrices creates a global list (DiscrimResults) for use in the JMP scripting language. The list contains a list of YNames, a list of XNames, a list of XValues, a matrix of YMeans, and a matrix of YPartialCov (covariances). An example from the iris data DiscrimResults is
Get Discrim Matrices, only available through scripting, obtains the same values as Save Discrim Matrices, but returns them to the caller rather than storing them in the data table.
Show Within Covariances shows or hides the Covariance Matrix report. The report for this example is shown in Covariance Matrix Report.
Covariance Matrix Report
Show Group Means shows or hides a table with the means of each variable. Means are shown for each level and for all levels of the X variable. Group Means Table shows the Group Means table for this example.
Group Means Table